The same director name appears on filings in three jurisdictions. The fields that would tell you whether it is one person, several, or a deliberate blur are scattered across registries that do not normalise the same way.
The problem
A name in a corporate filing is not an identifier. It is a label. Two filings with the same label may describe the same person, two unrelated people sharing a common surname, or one person whose name has been transliterated differently across registries that use different conventions. The discriminating fields — date of birth, nationality, address, prior directorships — are themselves recorded inconsistently across jurisdictions, with different formats, different romanisations, and different field-level completeness.
The standard manual approach is to open each registry record in a separate tab and compare fields by eye. This is fine for two records. It breaks at three or more. The analyst's attention drifts to the fields that match first, and the fields that should be doing the discriminating work — the ones with subtle format differences — get less scrutiny than they deserve. The error mode is not failing to spot a match. It is over-matching: collapsing distinct individuals into one identity because the name and one or two adjacent fields align.
What the analyst needs before opening a research tab is a structured matrix showing each registry extract normalised onto the same row, with the divergent fields flagged. That matrix is what the prompt in this piece produces.
This tutorial is for Signal subscribers.
Methods goes deep on a single technique each fortnight. The decision framework, the tools, the failure modes, and the evidentiary standard required to use the finding defensibly.
Join SignalA Signal subscription gives you:
- Full OSINT Reference Card library
- Methods, all tradecraft tutorials in full
- Shadow Analysis, all evidence-based reporting
- Forensic Dossiers, full access
- Discord access included
